Medicine is a field of uncertainties and of probability. Faced with questions like, “Doc, will this help?” or “Should I take this test?”, physicians must distill statistical concepts like the number needed to treat or the absolute risk reduction into easy-to-understand answers.
Not surprisingly, many find this difficult, according to a recent study published in JAMA Internal Medicine.
University of Maryland researchers gave 117 attending physicians and residents in internal medicine a short, written questionnaire assessing their knowledge of the benefits and harms of several common therapies or screening techniques. For example, one question asked about the frequency of a cancer diagnosis following a mammogram that screens “positive.” (The correct answer is between 1 and 4.9 percent).
On the questionnaire, most of the participants overestimated both the benefits of treatments — 79 percent average overestimate across all benefits questions — and the harms — 66 percent average overestimate across all harm questions. The survey did not examine guidance the doctors had actually provided to their own patients.
The respondents were aware of their lack of knowledge about the probability of benefit or harm — 67 percent rated their confidence level four or lower on a scale from one to 10, and most indicated they would be willing to use a website or other tool to help determine the correct statistics to provide for patients.
The results point out a need for increased emphasis on communicating probability and other relevant numerical information throughout medical school and residency, the authors write.